Vloop API
    Preparing search index...

    Interface Chat

    A chat object returned by ai. Use ai methods to create or find a chat.

    interface Chat {
        chatTag: string;
        providerSessionId: string | null;
        runner: AgentRunner;
        status: LoopStatus;
        title: string;
        vloopexId: string;
        cacheWarmerPingIntervalMs(): number | "auto";
        cacheWarmerPingIntervalMs(value: number | false | "auto"): Chat;
        childChats(): ChatRef[];
        commandOutput(): string;
        disableCacheWarmer(): Chat;
        done(): Promise<void>;
        enableCacheWarmer(setting?: number | "auto" | undefined): Chat;
        enough(): Chat;
        followup(options: ChatTurnOptions): Chat;
        hold(): Chat;
        incomplete(): boolean;
        on(eventName: string, handler: (data: object) => void): () => boolean;
        processStderr(): string;
        promptCache(): Promise<PromptCacheReport>;
        queue(options: Omit<ChatTurnOptions, "whenBusy">): Chat;
        readMedia(part: ReplyMediaPart): Promise<Buffer<ArrayBufferLike>>;
        ref(): ChatRef;
        releaseHold(): Chat;
        reply(): string;
        replyContent(): ReplyContentPart[];
        saveMedia(part: ReplyMediaPart, file: string): Promise<string>;
        steer(
            options: Pick<
                ChatTurnOptions,
                "timeoutMs"
                | "input"
                | "prompt"
                | "promptFile",
            > & { force?: boolean },
        ): Chat;
        stop(
            options?: {
                force?: boolean;
                reason?: string;
                resolve?: boolean;
                terminal?: boolean;
            },
        ): Chat;
        transcript(): string;
        wait(): Promise<void>;
    }
    Index
    chatTag providerSessionId runner status title vloopexId
    cacheWarmerPingIntervalMs childChats commandOutput disableCacheWarmer done enableCacheWarmer enough followup hold incomplete on processStderr promptCache queue readMedia ref releaseHold reply replyContent saveMedia steer stop transcript wait
    chatTag: string

    Saved chat tag, or an empty string for an unnamed chat.

    providerSessionId: string | null

    Raw provider session ID, when one is known.

    runner: AgentRunner

    Runner used by this chat.

    status: LoopStatus

    Status of the turn that runs or ran last, including a cache warm-up.

    title: string

    Current chat title; follows renames.

    vloopexId: string

    Stable Vloopex chat ID, also shown in the workbench Details.

    • Gets the cache warmer setting: "auto", a number of milliseconds, or 0 when the warmer is off. Warm-ups run on the Vloopex server.

      Returns number | "auto"

    • Sets the cache warmer. "auto" warms before the cache retention ends. A positive interval warms that many milliseconds after each cache use. Zero or false disables warming. A negative number throws. The server applies the change at once. A chat whose provider runs in the loop process keeps the setting but does not warm.

      Parameters

      • value: number | false | "auto"

      Returns Chat

    • Turns the cache warmer of this chat off: no more warm-ups are scheduled. A warm-up that runs ends as a turn does; stop() stops it.

      Returns Chat

    • Waits for the loop's work: the turn that the latest prompt of the loop or of a user started, and the prompts that the loop queued after it. Failed turns reject. A user-paused chat keeps waiting until a later turn finishes. A cache warm-up alone is not loop work: it neither delays nor settles done().

      Returns Promise<void>

    • Turns the cache warmer of this chat on.

      Parameters

      • setting: number | "auto" | undefined = "auto"

        "auto" (the default) warms before the cache retention ends; a positive number of ms warms that long after each use of the cache.

      Returns Chat

    • Starts another turn. While another turn of the chat runs, for example one that a user or a cache warm-up started, whenBusy decides what the prompt does, as the workbench offers Queue and Steer then. The owner of the chat decides with the state that it has when the follow-up arrives: the server for a chat that it owns, this chat object for a provider that runs in the loop process.

      Parameters

      Returns Chat

    • Keeps done() pending after each reply, also a failed or timed-out one, until releaseHold() or enough(). Sends, steers, queued prompts, and stops keep the hold; a terminal stop ends it. docs/lifecycle.md gives the rules.

      Returns Chat

    • Reports whether the chat result may be partial because its turn did not finish successfully, or is not there because the server lost it.

      Returns boolean

    • Runs a handler for matching chat events.

      Parameters

      • eventName: string

        Event name to observe.

      • handler: (data: object) => void

        Function called with event data.

      Returns () => boolean

      Function that removes this handler.

    • Returns string

      Raw process error output collected from the runner.

    • Sends a prompt after the current turn finishes successfully, or starts it immediately when the chat is idle. A queue that the user paused in the workbench sends it after Resume queue.

      Uses resumeChoice and forkSource when the queued turn starts.

      Parameters

      Returns Chat

    • Reads the bytes of a saved or inline binary part from this chat. Provider links are not downloaded. An unreadable saved resource rejects the read; inline data does not replace it.

      Parameters

      Returns Promise<Buffer<ArrayBufferLike>>

      The media bytes.

    • Ends the hold. done() then resolves or rejects from the reply that the hold kept, as without a hold, unless another pause, such as a Stop, keeps it. docs/lifecycle.md gives the rules.

      Returns Chat

    • Returns string

      Model text of the chat result: the reply of the latest turn that a prompt of the loop or of a user in Vloopex started, and before such a turn, the latest reply in the provider history. A cache warm-up alone does not replace it.

    • Saves a media part without changing its bytes. Creates parent folders and rejects if the destination already exists. Does not convert formats.

      Parameters

      • part: ReplyMediaPart

        A media part from replyContent().

      • file: string

        Destination path, relative to the project or absolute.

      Returns Promise<string>

      The absolute path of the saved file.

    • Interrupts a running turn and sends a replacement prompt. When the chat is idle, starts a normal follow-up.

      Parameters

      • options: Pick<ChatTurnOptions, "timeoutMs" | "input" | "prompt" | "promptFile"> & {
            force?: boolean;
        }

        force defaults to false when the driver supports steering during a turn, and true otherwise. A forced steer interrupts the turn before sending.

      Returns Chat

    • Stops the current turn. With no options this pauses the chat, so done() stays pending until a later turn finishes.

      Parameters

      • Optionaloptions: { force?: boolean; reason?: string; resolve?: boolean; terminal?: boolean } = {}
        • Optionalforce?: boolean

          Forcefully stops the process. Defaults to false.

        • Optionalreason?: string

          Text included if stopping becomes an error.

        • Optionalresolve?: boolean

          Accepts the partial reply and lets done() finish. Prefer enough() in normal loop code. Defaults to false.

        • Optionalterminal?: boolean

          Ends the wait as a final abort instead of a pause, also while Hold is on, and ends the hold. Used mainly by runtime cleanup. Defaults to false.

      Returns Chat

    • Returns string

      All prompts and output of the chat, cache warm-ups included.

    Settings
    Member Visibility

    On This Page
    Properties
    Methods